Emerging resistance among bacterial pathogens in the intensive care unit – a European and North American Surveillance study (2000–2002)

Background Globally ICUs are encountering emergence and spread of antibiotic-resistant pathogens and for some pathogens there are few therapeutic options available. Methods Antibiotic in vitro susceptibility data of predominant ICU pathogens during 2000–2 were analyzed using data from The Surveillance Network (TSN) Databases in Europe (France, Germany and Italy), Canada, and the United States (US). Results Oxacillin resistance rates among Staphylococcus aureus isolates ranged from 19.7% to 59.4%. Penicillin resistance rates among Streptococcus pneumoniae varied from 2.0% in Germany to as high as 20.2% in the US; however, ceftriaxone resistance rates were comparably lower, ranging from 0% in Germany to 3.4% in Italy. Vancomycin resistance rates among Enterococcus faecalis were ≤ 4.5%; however, among Enterococcus faecium vancomycin resistance rates were more frequent ranging from 0.8% in France to 76.3% in the United States. Putative rates of extended-spectrum β-lactamase (ESBL) production among Enterobacteriaceae were low, <6% among Escherichia coli in the five countries studied. Ceftriaxone resistance rates were generally lower than or similar to piperacillin-tazobactam for most of the Enterobacteriaceae species examined. Fluoroquinolone resistance rates were generally higher for E. coli (6.5% – 13.9%), Proteus mirabilis (0–34.7%), and Morganella morganii (1.6–20.7%) than other Enterobacteriaceae spp (1.5–21.3%). P. aeruginosa demonstrated marked variation in β-lactam resistance rates among countries. Imipenem was the most active compound tested against Acinetobacter spp., based on resistance rates. Conclusion There was a wide distribution in resistance patterns among the five countries. Compared with other countries, Italy showed the highest resistance rates to all the organisms with the exception of Enterococcus spp., which were highest in the US. This data highlights the differences in resistance encountered in intensive care units in Europe and North America and the need to determine current local resistance patterns by which to guide empiric antimicrobial therapy for intensive care infections.


Background
Antimicrobial resistance has emerged as an important factor in predicting outcomes and overall resource use after infections in intensive care units (ICU) [1]. Globally ICUs are encountering emergence and spread of antibioticresistant pathogens. For some pathogens there are few therapeutic options available, e.g., vancomycin-resistant Enterococcus faecium. Awareness of these problems has been underscored with data from a number of surveillance studies aimed at improving the use of empiric therapy. In the United States there have been several national programs, which have focused on both the etiology of infections and resistance patterns of nosocomial or ICU infections including the National Nosocomial Infections Surveillance (NNIS) [2] and more recently an ICU-specific study examining the epidemiology of antimicrobial resistance, Project ICARE [3,4]. Stephen et al. collected strains from 28 ICUs from across the United States as part of the SENTRY Antimicrobial Surveillance Program in 2001 [5].
European data on the antimicrobial resistance of ICU pathogens has also been collected in several recent surveillance studies. A large prevalence survey of nosocomial infections in ICUs in 17 countries was published in 1995 [6], and more recently a number of nation-specific surveys were reported [7][8][9]. Several key points emerge: first, antimicrobial resistance among ICU pathogens is generally increasing, but variations do exist among different countries, probably due to individual antimicrobial use patterns; second, when new medical practices and alternative antimicrobials are introduced changes in the dominant microbial etiologies may emerge prompting novel empiric selections; and third, the standards of hygiene and infection control also vary across countries. Finally, appropriate therapy of ICU infections directed by local resistance data can have significant consequences for both patient and the healthcare system. It is against this background that local resistance surveillance programs are of most value in developing appropriate therapeutic guidelines for specific infections and patient types. For example, the recent modification to the American Thoracic Society guidelines for the treatment of hospital-acquired pneumonia [10] considered contemporary resistance data. Local surveillance data can be applied to other infections to assist in local formulary policy such as those governing treatment of nosocomial urinary tract infections [11].
This study using TSN program reports the antimicrobial resistance profiles of bacterial isolates from ICU patients in five countries during the period 2000-2002. The relevance of these recent nation-specific data will be discussed on a country-by-country basis, as part of improving and updating empiric therapeutic approaches to specific pathogens causing infections in the ICU setting according to each country. These surveillance programs help to maintain current knowledge of susceptibilities and relevant treatment options.

TSN Database -United States and Europe
TSN is a queriable, real-time database that electronically assimilates daily antimicrobial susceptibility testing and patient demographic data from a network of geographically dispersed laboratories in the United States (283 hospital sites), France (63 hospital sites), Germany (169 hospital sites), Italy (48 hospital sites) and Canada (87 hospital sites) [12].
Laboratories included in TSN include those servicing university, community, and private hospitals with bed sizes ranging from 100 to >1000 beds. Routine diagnostic susceptibility testing results are collected daily from each participating laboratory. The methods used by these laboratories include VITEK (bioMérieux, St. Louis, MO), MicroScan (Dade-Microscan, Sacramento, CA), Sceptor and Pasco MIC/ID (Becton Dickinson, Sparks, MD) and Etest (AB Biodisk, Solna, Sweden), as well as manual broth microdilution MIC, disk diffusion and agar dilution. TSN reflects current testing in participant laboratories and represents the data reported to physicians from the respective laboratories [13].
Although some European countries have alternate breakpoints, all data forwarded to TSN Databases are derived from hospitals that utilized NCCLS standards and definitions (United States, Canada, Italy, and Germany) [14] or the Comité de L'Antibiogramme de La Societé Français de Microbiologie (France) [15] [16]. In addition, a series of quality-control filters (i.e., critical rule sets) were used in TSN to screen susceptibility test results for patterns indicative of testing error and suspect results were removed from analysis for laboratory confirmation. In TSN, any result from the same patient with the same organism identification and the same susceptibility pattern received within five days was considered a repeat culture and was counted only once in the database.

Bacterial species and antimicrobials tested
For this study, data from TSN results for each individual database from January 1, 2000 through to December 31, 2002 were included in the analysis to determine the proportion of species and their susceptibility to antimicrobial agents commonly tested in clinical laboratories throughout the participating regions. Only isolates derived from patients located in hospital ICUs were considered in the analysis. Gram-positive species included in the analysis were comprised of S. aureus, coagulase negative staphylococci, Enterococcus faecalis,Enterococcus faecium, Streptococcus pyogenes, Streptococcus pneumoniae and viridans group streptococci. Gram-negative species studied comprised the predominantly encountered enteric species (Escherichia coli, Klebsiella oxytoca, Klebsiella pneumoniae, Proteus mirabilis, Morganella morganii and Serratia marcescens), and Pseudomonas aeruginosa and Acinetobacter spp.
The antibiotics studied are listed in Tables 2,3,4,5. Among E. coli, putative ESBL production was defined as those isolates that were intermediate or resistant (non-susceptible) to ceftazidime [17]. Given the large number of isolate results included in the majority of analyses in this study, statistical analysis was not performed, as even subtle differences in percent resistance (<1%) to an antimicrobial agent for any time period or demographic parameters would be reported as highly significant (P <0.001).

Results
In vitro susceptibility data from over 220,000 isolates from ICUs in five countries over the period 2000-2002 were assimilated. The most frequent species isolated from infections in the ICU was S. aureus, being most common in three of the five countries ( Table 1). The oxacillin resistance rates among S. aureus varied markedly across countries from 19.7% in Canada to 59.5% in Italy. E. coli (7.7%-15.5%) and P. aeruginosa (10.8%-22.3%) were the most frequent Gram-negative organisms encountered. The Gram-positive genus Enterococcus, either as E. faecalis, E. faecium or non-speciated isolates accounted for <10% of isolates in most countries with E. faecalis being the most common species <4.3%. Community-acquired respiratory pathogens such as Streptococcus pneumoniae and Haemophilus influenzae were relatively uncommon in all five countries. Tables 2,3,4,5 show the antimicrobial susceptibility profiles of various Gram-positive and Gram-negative pathogens isolated from ICU patients against a range of relevant antimicrobials.
Specifically notable susceptibility patterns include the vancomycin susceptibility of all strains of staphylococci. Generally, there was a low proportion of vancomycin resistant E. faecalis <5%, whereas vancomycin-resistant E. faecium was more prevalent ranging from 0.8% in France to 76.3% in the United States, with a wide inter-country variation ( Table 2). Penicillin resistance rates varied among S. pneumoniae, from 2.0% in Germany to 20.2% in the US with concurrent ceftriaxone resistance rates of 0% in Germany to 3.4% in Italy (Table 3).
β-lactam activity was assessed by comparing four different cephalosporins and a β-lactam/β-lactamase inhibitor combination, piperacillin-tazobactam. Overall, the putative production of ESBLs among E. coli was low, <6%, but ceftazidime resistance was reported at higher rates in K. pneumoniae and S. marcescens, with the highest rates seen in M. morganii, from 16.0% in Germany to 26.4% in the United States (Table 4). Among the gram-negative organisms tested, ceftriaxone resistance rates were usually lower than ceftazidime, with the exception among P. aeruginosa and Acinetobacter spp. Cefepime, a fourth generation cephalosporin with anti-pseudomonal activity was also more active than ceftazidime (Table 5). Against the Enterobacteriaceae, the β-lactam combination agent piperacillintazobactam was generally less active than ceftriaxone. These species showed a wide variation in fluoroquinolone susceptibility among both species and countries. Gentamicin resistance rates among the Enterobacteriaceae varied from 1.2% among K. oxytoca from Germany to 37.2% in P. mirabilis from Italy. Ciprofloxacin resistance rates among E. coli ranged from 6.5% in France to 12.7% in

Discussion
Data derived from international surveillance studies, such as those presented here, can provide a unique contemporary perspective on the susceptibility of commonly encountered organisms to commonly used antibiotics. Such surveillance systems play a crucial role in detecting emerging trends in resistance. Comparisons of these with data of other recent surveillance programs show the wide variations in susceptibility profiles and the need for ongoing unit-specific surveys.
In Germany the prevalence of resistance among gram-positive organisms remained comparatively low with an incidence of 21% MRSA.     1994-98 to 1999, but only a 2% increase among CNS strains [4]. Vancomycin resistance in the United States was observed in 4.5% of E. faecalis; however, over 76% E. faecium were vancomycin non-susceptible.
Although streptococci are uncommon ICU pathogens they can be rapidly invasive and possibly fatal unless adequate therapeutic approaches are adopted. S. pneumoniae in the United States has acquired a range of resistance mechanisms with resistance to penicillin and the macrolides, clarithromycin and erythromycin, being common, 20.2% and 25.5%-30.5% respectively. The newer generation cephalosporins, ceftriaxone, cefotaxime and cefepime showed good activity against pneumococci, 3.2%, 6.3% and 4.5% resistant, respectively. Less than 1.0% of isolates were resistant to levofloxacin. These data are similar to other recent reports [20].
For Enterobacteriaceae which account for approximately 30% of all isolates from ICU infections, the incidence of putative ESBLs was low in E. coli, 4.7% but ceftazidime non-susceptibility was higher in K. oxytoca 8.3%,K. pneumoniae 11.5%,S. marcescens 10.3% and M. morganii 26.4%. These data are consistent with other recent reports [21]. Fluoroquinolone resistance was observed in all Enterobacteriaceae tested, in the US for example, resistance rates were as follows, using ciprofloxacin as a marker: E. coli 10.7%, K. oxytoca 5.9%, K. pneumoniae 8.4%, M. morganii 20.7%, P. mirabilis 12.7% and S. marcescens 6.7%. These data show increased fluoroquinolone resistance compared with recent reports [21]. Jones et al. previously reported susceptibility data on ICU pathogens isolated over the period 1998-2001 [22].
Specifically, enteric bacteria showed changes over this time. Fluoroquinolone resistance doubled among E. coli isolates from 3.3-5.5% to 10.8-11.4% [22]. This study showed a generally higher level of activity among thirdgeneration cephalosporins than other reports [23], with ceftriaxone showing <10% resistance rates against most species tested. Piperacillin-tazobactam showed less consistent activity with some species being >14% resistant, e.g. Klebsiella spp.,P. aeruginosa, and Acinetobacter spp. present significant therapeutic challenges in ICUs in the United States. With the exception of cefepime, all other tested antimicrobials demonstrated >12% resistance to P. aeruginosa, many considerably higher. Piperacillin-tazobactam showed the next lowest resistance rate, 14.4%, with all other agents having rates of 17% or higher. Nonsusceptibility to ciprofloxacin among P. aeruginosa was 37.2%, higher than in the Neuberger report. Sahm et al. reported a 10% increase in fluoroquinolone resistance among P. aeruginosa in the United States, whereas resistance emerged more slowly with the other classes of antimicrobials tested [12]. Acinetobacter infections continue to present significant therapeutic challenges due to the extensive resistance mechanisms demonstrated by the >25% resistance shown in Table 5. Only imipenem has any reliable activity against Acinetobacter spp. with an 87% susceptibility rate.
There are several implications of these data. It is essential that local surveillance programs be maintained in each country's ICU setting. The local data are vital to the formulary committees as they select appropriate agents to treat infections. There are clear differences among the five countries studied in this report. Although the predominant pathogens are similar, ongoing surveillance is essential to detect the emergence of resistant species. It is clear that certain classes of compounds are losing activity against the ICU pathogens tested. For example, the fluoroquinolones have reduced susceptibility among many Gram-negative species as well as staphylococci; however, the newer class members have enhanced activity against pneumococci. Advanced-generation cephalosporins have variable activity, with ceftriaxone showing consistently good activity against the Enterobacteriaceae and some staphylococci. Ceftazidime has lost potency due to the emergence of ESBL enzymes and also has diminished activity against P. aeruginosa. Piperacillin-tazobactam is generally active against P. aeruginosa in ICUs. The aminoglycoside, gentamicin has shown continued activity against most Enterobacteriaceae in all five countries, and modest activity against S. aureus but not against CNS strains. The gentamicin susceptibility of P. aeruginosa ranged from 44.0% in France to 74.0% in Germany, whereas Acinetobacter spp . showed more variable gentamicin susceptibility varying from 23.3% in Italy to 82.0% in Germany. These local data should be considered when treating infections in the ICU.
Use of agents with anti-pseudomonal activity such as cefepime, piperacillin-tazobactam or the carbapenems should preferably be reserved for patient types or infections where this pathogen is present or risk factors exist, as per the ATS Community acquired-pneumonia guidelines [24]. A combination of a third-generation cephalosporin such as ceftriaxone with vancomycin may be appropriate for bloodstream infections based upon the NNIS etiology data from 1992-1999.

Conclusions
The current study confirmed the emergence of fluoroquinolone resistance among various Gram-negative species and staphylococci, which may be increasing due to the heightened use of these drugs; however the reported ESBL rates among Enterobacteriaceae was lower than noted in other studies and appeared to be stable. The prevalence of MRSA, perhaps the most significant resistant hospital pathogen, varied among the five countries and appeared to be increasing. Parenteral cephalosporins such as ceftriaxone and cefotaxime remained quite active against Enterobacteriaceae. Up-to-date susceptibility data should be made available as rapidly as possible to physicians so that appropriate targeted empirical therapy can be instituted, this approach can assist in maintaining the activity of the current antimicrobials. While local surveillance studies remain crucial, national surveillance studies such as this can provide an invaluable data source to provide guidance in formulary decision-making.

Authors Contributions
MJ conceived the study, provided data interpretation and drafted the manuscript. DD analyzed the study data; JK and DS provided expert microbiological analysis and interpretation of study data; RW provided clinical expertise in interpretation of data and drafting manuscript. All authors read and approved the final manuscript.